Characterization of the Molecular Landscape of Pediatric Acute Myeloid Leukemia: A Retrospective AIEOP AML2013/01 Study

Background. Advances in genomic techniques have recently shed light on leukemia biology, recognizing that pediatric acute myeloid leukemia (AML) is a highly genetically heterogeneous disease. In the last two decades, genetic and cytogenetic abnormalities contributed to the stratification of patients in different classes of risk, where children receive a risk-adapted therapy. In the recently concluded AIEOP-AML2013/01 clinical trial, 371 Italian pediatric AML patients were allocated into three risk groups that were largely based on measurable residual disease at the end of induction therapy and on recurrent molecular markers with a recognized impact on children’s outcome.

Methods and Results. We screened by RT-PCR, RQ-PCR, capillary electrophoresis and Sanger sequencing 41 genetic markers on 371 RNA and DNA samples, identifying a molecular marker at diagnosis for 247 cases out of the 371 enrolled (67%). As far as gene mutations detected are concerned, we identified 24/371 cases harboring a mutation in exon 12 of NPM1 gene (6.5%), 16/371 a FLT3-ITD mutation (4.3%), and 7 harboring both NPM1 and FLT3-ITD mutations (2%). Moreover, we also found 2 cases having a partial tandem duplication affecting KMT2A gene (2/371, 0.5%), one of them occurring together with a FLT3-ITD mutation. For chromosomal aberrations, we confirmed that t(8;21)RUNX1::RUNX1T1 and inv(16)CBFB::MYH11 are recurrent translocations in pediatric AML, being found in 13.5% (50/371) and 12% (44/371) of cases respectively. Sixty-four (17,3%) AML cases were characterized by a KMT2A fusion with different partner genes, with the t(9;11)KMT2A::MLLT3 being the most recurrent (24/371, 6.5%), followed by t(10;11)KMT2A::MLLT10 (21/371, 5.7%), t(6;11)KMT2A::AFDN (10/371, 2.7%), t(11;19)KMT2A::ELL (6/371, 1.6%). In 14 patients, we found a NUP98 translocations (3.8%), and, in particular, 9 cases harbored the t(5;11)NUP98::NSD1 (2.4%, 5 of them together with a FLT3-ITD mutation), 4 the t(11;12)NUP98::KDM5A (1%) and only one case the t(2;11)NUP98::HOXD13 (0.3%). Other chromosomal aberrations were rarer in our cohort, with t(16;16)CBFA2T3::GLIS2 accounting for 2.4% of cases (9/371), t(6;9)DEK::CAN 1.3% (5/371, co-occurring with FLT3-ITD in 3/5 cases), t(1;22)RBM15::MKL1 1.1% (4/371), t(3;5)NPM1::MLF1 0.5% (2/371, one concomitant with FLT3-ITD), t(16;21)FUS::ERG 0.5% (2/371), t(16;21)RUNX1::CBFA2T3 0.3% (1/371), del(9)SET::NUP214 0.3% (1/371), t(8;16)KAT6A::CREBBP 0.3% (1/371), t(9;22)BCR::ABL1 0.3% (1/371). In addition, we retrospectively evaluated the incidence and prognostic relevance of mutations at specific loci of KIT, FLT3, ASXL1, WT1 and CEBPA genes. As concerns the KIT gene, we identified mutations at exon 8 or 17 in 16/371 AML (4.3%). Of note, all the 16 cases had a concomitant core-binding factor (CBF) rearrangement, and in particular 10/16 the t(8;21)RUNX1::RUNX1T1 translocation and 6/16 the inv(16)CBFB::MYH11, with a relative occurrence of 17% in the CBF-r AML subgroup (16 KIT mutated AML/94 CBF-r) and a slight, but not significant, impact on outcome. As regards exon 13 of ASXL1 gene and the tyrosine-kinase domain of FLT3 gene (FLT3-TKD), our study confirmed a low prevalence of these mutations, which occurred in 2.5% and 3.4% of AML respectively. On the contrary, WT1 mutations (exons 6-9) occurred in a consistent portion of AML (45 mutated cases, 13.2%)did not affect the outcome of FLT3-ITD mutated AML. Conversely, by screening both at bZIP and NTD domains of the CEBPA gene, we identified 18 double-mutated patients (4.8%), and notably all these patients had no other molecular marker detected at diagnosis, reducing the proportion of negative cases from 33% to 28%. We found that CEBPA mutated cases are characterized by a favorable outcome (EFS 93% vs 58% for CEBPA wt, p=0.008), this finding corroborating the indication to allocate these patients to the standard risk group.

Conclusion. Overall, in the prospective AIEOP AML2013/01 trial, we described the frequency of different molecular lesions and improved the genetic landscape of AML. The future use of NGS-based screenings could further optimize the characterization of the genetic landscape of childhood AML, thus refining the risk class patients stratification and modulation of treatment approaches.